CSV & Parquet file validator
Drop a CSV/TSV to auto-detect its delimiter, quote, header and column types — and catch when the delimiter is wrong — or a Parquet file to check its magic bytes and footer for truncation or corruption. Everything runs in your browser (DuckDB-WASM); nothing is uploaded.
Drop a file or click to browse
Parquet or CSV/TSV — checked locally in your browser, never uploaded
Frequently asked questions
- Why won't my CSV load correctly?
- Usually a wrong delimiter, an unquoted comma, or ragged rows (lines with too many or too few fields). This tool shows the delimiter DuckDB detects and how many columns it finds — if it collapses to a single column, the delimiter is wrong or the rows are inconsistent, which is the usual culprit.
- How does it detect the CSV delimiter?
- It uses DuckDB's CSV sniffer (sniff_csv) to infer the delimiter, quote character, escape character, header row and rows to skip, then reports each column and its inferred type — the same detection DuckDB uses when it reads a CSV.
- What does Parquet validation check?
- It verifies the leading and trailing PAR1 magic bytes (a missing trailing magic is the classic sign of a truncated or still-uploading file) and confirms the footer/metadata reads cleanly, reporting the row and row-group counts and the writer.
- Can it repair a corrupt file?
- No — and be wary of anything that claims to. A corrupt Parquet footer generally isn't recoverable in the browser (or anywhere without the original writer). This tool diagnoses the problem so you can fix it at the source.
- Is my file uploaded?
- No. All checks run in your browser with DuckDB-WASM; the file never leaves your device.